Machine Vision
Deep learning on embedded platforms
At Automatica, MVTec will be focusing on the latest versions of its Halcon and Merlic software products. Live demos will demonstrate the use of modern machine vision technologies in automation scenarios. There will also be insights into new image processing functions relating to parallelization and matching.
At the trade fair stand, experts will demonstrate the use of complex deep learning algorithms on embedded platforms: Using the example of the Jetson TX2 embedded board from Nvidia, it will become clear how various objects such as tablets or fruit can be classified using deep learning. Characters and number combinations are also reliably recognized as part of OCR applications.
Another demo visualizes an application scenario from robotics: a robotic arm reaches into a collection of objects and, thanks to modern matching technologies from MVTec Halcon, accurately finds the position of the relevant object. The arm removes the object precisely from the box, recognizes it using a 2D camera and deep learning technology and sorts it out.
Functions of the current preview version of Merlic 4 will also be illustrated in a demo: two cameras will be used to solve various inspection tasks and features relating to parallelization, i.e. the parallel execution of independent tools, will be demonstrated. It will also be shown how Merlic uses deep learning-based OCR technologies to recognize different font types on packaging, such as best-before dates or batch numbers. The seamless integration of a programmable logic controller (PLC) into vision systems with Merlic will also be demonstrated. In the future, Merlic will also be integrated even better into automation solutions via Hilscher cards, for example with the help of Profibus. The first successful tests have already been carried out for this development. as

Interaktives Deep Learning
MVTec stellt seine die beiden Flaggschiff-Produkte Halcon und Merlic in den Fokus. Einen Schwerpunkt bilden dabei innovative Deep-Learning-Funktionen auf Basis von künstlicher Intelligenz.
Hall B5, Stand 305









